Interpreting Predictive Probabilities: Model Confidence or Human Label Variation?

Open in new window